Empirical literature has used the estimates provided by the Business Software Alliance to explain the phenomenon of software piracy. One measure that is present in all the studies is the Gross Domestic Product per capita (GDPpc). Several approaches were used: surveys using respondents from universities and in the labor market; longitudinal/panel studies and cross sectional studies; the last two rely on macroeconomic data. Results presented by these studies are very important complementing each other and, at the same time, they provide actions for policymakers.
Empirical literature that uses surveys can obtain richer results, being able to model each parameter (age, sex, income), but it relies on the willingness of the respondents to answer truthfully. Even if the inquiry is anonymous, due to the nature of the crime, they may sometimes underestimate responses. Surveys are used in a particular group of the population (students, business users) in a particular city. Many questionnaires rely on a likert scale6. When respondents answer questions it is possible that they go to the extremes or the middle (neither agree nor disagree), which can be sometimes a problem. In 2010 Business Software Alliance, with the help of IPSOS, performed a survey on 15000 computer users7 to measure the commercial value of unlicensed software and the piracy rates.
When surveys are implemented they suffer from a population bias problem, which can influence the main findings and extension of results. These studies covered specific population, like students Ram D. Gopal and Sanders (1998), Butt (2006), Higgin (2006) and Gan and Koh (2006) or business users Lau (2004). To overcome these problems authors such as Ram D. Gopal and Sanders (1998) and Holm (2003) used a cross sectional model that explained the phenomenon at a country level, complementing the results from the surveys.
Several factors can influence questionnaires, from the group of people surveyed, to the age, sex and location of the survey. Among the questions that can be asked we can find the following:
6 A Likert scale is a psychometric scale commonly involved in research that employs questionnaires. It is the most widely used approach to scaling responses in survey research, such that the term is often used interchangeably with rating scale. Usually it is divided into 5 ordinal values: 1. Strongly disagree, 2. Disagree;
3. Neither agree nor disagree; 4. Agree and 5. Strongly agree. See Wuensch, Karl L. (October 4, 2005). "What is a Likert Scale? and How Do You Pronounce 'Likert?
7 For more information see http://portal.bsa.org/globalpiracy2010/
18
- Do you use pirated software and how often do you use it?
- Do you use legal, illegal or open source software?
- Income plays an important factor in the choice to pirate?
- Culture, education or legal system plays an important factor in this decision?
These four examples can measure simultaneously several influences that the cross-sectional or panel data analyses can lose. The location in which the survey is made can affect results. Lau (2004) conducted a survey in Hong Kong, which is a place with one of the highest piracy rates compared with the Western Europe (+33%), North America (+21%) and the European Union (+35%); in 2010 the piracy rate in Hong Kong was 45%. The main conclusion of this study is that knowledge of software copyright law and the availability of original software have direct effects on self-reported leniency towards software piracy.
Being the empirical literature an important source for both policymakers and researchers, but being at the same time still in it’s infancy, we compile the major macroeconomic findings found by previous authors. Several dimensions have been found to affect piracy: Economic, Cultural, Educational, Technological and Legal dimensions; these will be discussed on the next subsections.
Economic dimensions
Stylized fact 1: Gross Domestic Product per capita affects negatively software piracy and Gross Domestic Product Growth is influenced by the correct enforcement of Intellectual
Property Rights.
Income affects the decision to purchase or to pirate by the consumers or firms. One measure that is present in many studies on the determinants of software piracy is the Gross Domestic Product per capita. Some examples are Ram D. Gopal and Sanders (1998), Marron and Steel (2000) and Goel and Nelson (2009). The results show that an increase in income can decrease software piracy. Other measures can be used that reflect the levels of income of a country; Holm (2003) used the Gross National Income per capita (GNIpc) and obtained the same results. Levels of income are heterogeneous among countries,
19 furthermore, many software products are sold at the same price across countries; examples are movies, video games and music. Shin, Gopal, Sanders, and Whinston (2004) split the GDPpc into two subsamples: one which represents income less than 6 000$ and other that represents more than 6 000$. In countries that have GDPpc less than 6 000$, income affects negatively software piracy (-0.0032), but when GDPpc is higher than 6 000$, this negative effect becomes marginal (-0.0008). This result indicate that on households that have more disposable income the fraction of the income that is allocated to software is reduced. On the other hand, when the income is low this fraction increases. Increasing income on households with less income will result in less software piracy.
Other authors studied what were the effects of piracy on economic growth. In spite of high piracy rates, indicating that property rights protection were not perfect, Andrés and Goel (2012) found that the existence of software piracy increased economic growth. Using an index of Intellectual property Rights, Park and Ginarte (1997) and Falvey, Foster, and Greenaway (2006), found that intellectual property rights could promote growth.
Stylized fact 2: Income inequality measured by the GINI index affects negatively software piracy.
Additional work was done in explaining these differences using the GINI Index. To check this, Fischer and Andrés (2005) used a sample of 71 countries to analyze the relationship between income distribution and software piracy rates. To analyze this income inequality it is used quintile shares. This quintile analysis is divided into three classes: Q1 is low-income class; Q2-Q4 is middle-income class and Q5 is upper-income class. Software piracy is a middle class crime in Latin America, Caribbean, East Asia and the Pacific Regions. Software piracy is a crime committed by middle and lower class in the Central Asia and Eastern Europe and is an upper class crime in Western Europe and North America. In a recent study and using a sample of 35 countries, Andrés (2006b) found income inequality to be negatively related with software piracy; more equal societies have higher piracy rates.
In a theoretical paper Poddar (2005) tried to study differences of software piracy across countries; using the same variables of interest (GINI index), but with opposite results.
Poddar (2005) developed a model that assumes that software firms undertake R&D to prevent piracy, which can be replicated with measures of IPR (Intellectual Property Rights
20
protection). He considers three types of consumers: one that buys, one that pirates and other that do not use any type of software. These consumers are a simplification of the reality; in real life each one can, at the same time, use both legal and illegal software. A high income gap between users and a low protection cannot prevent software piracy. When this gap is reduced and with the existence of some protection, there is a probability of mitigating software piracy. This result was studied by Fischer and Andrés (2005) and Andrés (2006b) using the GINI index8.
Stylized fact 3: HDI affects positively software piracy
Software piracy can affect the development of a country; software development and distribution activities gives jobs to thousands of people, but these jobs are not necessarily made available where we buy the software. It can happen that national companies outsource software development to countries with highly qualified labor force but with lower wages.
Using a panel data combining three years (1995, 2000 and 2002), Bezmen and Depken (2005) study this phenomenon. The measure of economic development is introduced with the HDI (Human Development Index). They used an equation system. In the first equation, piracy rates were the dependent variable and, in the second HDI where the dependent variable. This measure was used by Boyce (2011) introducing GINI index as well. In both works this variable increase software piracy rates.
Cultural dimensions
Stylized fact 4: Hofstede cultural dimensions explain levels of software piracy across countries.
8 This variable “measures the extent to which the distribution of income among individuals, within an economy, deviates from a perfectly equal distribution”. A low value of this index represents an equal society while a high value represents an extremely unequal society. Source: Key Indicators of the Labour Market (KILM):2001-2002, International Labour Organization, Geneva, (KILM):2001-2002, page 704.
21 The Hofstede cultural dimensions (see G. Hofstede (2004)) cover several dimensions: power distance (PDI)9, individualism (IDV), uncertainty avoidance (UAI)10 and masculinity (MAS)11. They represent “four anthropological problem areas that different national societies handle differently: ways of coping with inequality, ways of coping with uncertainty, the relationship of the individual with her or his primary group, and the emotional implications of having been born as a girl or as a boy”12. They allow a comparative analysis between the national culture and the levels of software piracy. Although this measure allows a rich analysis, but suffers some drawbacks as it does not vary over time, and the sample covered is not large enough. In 1991 it was introduced a fifth dimension: the Long-Term Orientation (LTO)13. This dimension was developed by Minkov (2007). More recently, in 2010, it was introduced a sixth dimension: the Indulgence versus Restraint (IVR)14, developed by Geert Hofstede, Hofstede, and Minkov (2010).
Nevertheless, several authors used these dimensions to explain the levels of software piracy rates across countries. Some examples are Marron and Steel (2000), Moores (2003), Shin et al. (2004)15 and Kovačić (2007). These studies used a cross sectional analysis, covering at most 72 observations. Results show that individualism is negative and significant. Additional to this, Masculinity has a negative value and power distance a positive value. Other studies analyzed the effect of religion on the decision to pirate. Al-Rafee and Rouibah (2010) found that religion factors affect the decision to pirate. This was done with a questionnaire saying that, based on the individual religion, software piracy was stealing.
9 This dimension expresses the degree to which the less powerful members of a society accept and expect that power is distributed unequally
10 The uncertainty avoidance dimension expresses the degree to which the members of a society feel uncomfortable with uncertainty and ambiguity.
11 The masculinity side of this dimension represents a preference in society for achievement, heroism, assertiveness and material reward for success.
12 http://www.geerthofstede.nl/
13 The long-term orientation dimension can be interpreted as dealing with society’s search for virtue.
14 Indulgence stands for a society that allows relatively free gratification of basic and natural human drives related to enjoying life and having fun.
15 These authors used collectivism, which is the opposite of individualism. The high side of this dimension, called Individualism, can be defined as a preference for a loosely-knit social framework in which individuals are expected to take care of themselves and their immediate families only. Its opposite, Collectivism, represents a preference for a tightly-knit framework in society in which individuals can expect their relatives or members of a particular in-group to look after them in exchange for unquestioning loyalty. A society's position on this dimension is reflected in whether people’s self-image is defined in terms of “I” or “we.”
22
Educational dimensions
Stylized fact 5: Overall level of Education affects negatively the levels of software piracy.
Education plays an important factor in the construction of the perception of an individual towards using or not legal or illegal software. Several questions are raised with this respect: (i) more education can affect the levels of software piracy? ; (ii) education can bring an increase use of legal, illegal or both types of software? Several dimensions related to education can be used, from the literacy rate to the level of education attained. A challenge is posed on the availability of data for large group of countries. The World Bank, namely the World Development Indicators (WDI) has information on several dimensions related to education from the school enrolment ratio (primary, secondary, and tertiary), expenditure on education and years of primary and secondary schooling. The Eurostat provides a broader picture, introducing additional financial and not financial measures, but information is only available for a small group of countries (the European Union).
In spite of a broad range of variables available in this dimension, but due to data restrictions, cross-sectional analysis has been implemented restricting the analysis. This dimension has been studied by Marron and Steel (2000) and Andrés (2006b) with the introduction of average years of secondary education of people with more than 25 years old (Barro & Lee, 2013). Their results show that more education reduces software piracy. Goel and Nelson (2009) and Andrés and Goel (2011) used literacy rate; this variable has a positive sign. The statistical significance of this variable in the first study was at most 5% but, in the second study, significance was not achieved. Literacy rate omits the level of education attained; a person can be literate and have a low level of education. It also omits the various ISCED (International Standard Classification of education) levels. Measures that reflect the specific level attained by person measured by the ISCED 1997 or ISCED 2011 classification, reflect the expenditure on education and can improve results. Other measure that has been studied by MacDonald and Fougere (2003) is the inclusion of the word “software piracy” in textbooks. For this purpose he analyzes the MIS textbooks. Software piracy is present on 72% of the textbooks; Ethics is present in 67%, software license in 50%, copyright (50%) and Intellectual Property 39%. This is only an example of a particular field of knowledge;
23 introduction of additional fields of knowledge such as Management and Economics could improve results.
Technological dimensions
Stylized fact 6: Types of software protection affects levels of software piracy.
Choice of type of Internet access and associated services will depend on its price, availability and the utility given by additional services, which will affect the availability of
software.
Before the rise of the Internet, software piracy was made with the replication of the original software, from its original support, to several pirated CDs or floppy-disks;
protection was both in the software itself in the form of serial keys, some with many digits, and requiring a special number that was provided by telephone as an additional protection barrier. The hardware protection in PC software is generally attached to the support (CD, Floppy, etc.) and not in the PC itself; functional copies were more difficult to produce. It is often hacked with more or less effort.
There are different ways to protect software; some of these are License Keys and Product Activation16 (Anckaert, Sutter, & Bosschere, 2004). Djekic and Loebbecke (2007) studies the influence of technical copy protections on application software piracy, following Ram D. Gopal and Sanders (1997), Prasad and Mahajan (2003) and Anckaert et al. (2004), they distinguish between software-based and hardware-based technical copy protections. A survey is conducted using 219 professional users and an amateur group. Software based protection and hardware based protection are analyzed separately.
Personal context variables are always significant and positive. This context is represented by income, requirements of usage in the workplace and the intensity of
16 One example is the Windows Genuine Advantage (WGA). It is an anti-piracy system created by Microsoft that enforces online validation of the licensing of several recent Microsoft Windows operating systems when accessing several services, such as Windows Update, and downloading Windows components from the Microsoft Download Center. In Windows 7, WGA is renamed Windows Activation Technology. WGA consists of two components: an installable component called WGA Notifications that hooks into Winlogon and validates the Windows license upon each logon and an ActiveX control that checks the validity of the Windows license when downloading certain updates from the Microsoft Download Center or Windows Update.
http://en.wikipedia.org/wiki/Windows_Genuine_Advantage
24
application software usage. These variables affect more the amateur group. Legal software use that is protected with license key or product activation is higher in the amateur group while software that is protected with hardware protection is higher in the professional group.
This work shows that being able to work properly with software can affect their valuation of the software; the full capabilities and price of the software are understood. Some productivity tools like Photoshop® can be pirated by home users but the full capabilities are not used.
This can be seen by the firm as a loss, but this might not be completely true if we consider an inexperienced user. On the other hand if this software is used illegally at the workplace this is not true; it is a loss, the worker knows how to use the software at its full.
When the hardware protection and Software protection is overcome by hackers, the next step is to upload the software, which will depend on the type of Internet access and upload speeds. Hackers may use public Internet providers such as universities. Broadband Internet access plays an important role in the decisions to download legal or illegal software by potential pirates. One of the first studies in Europe that focuses on the demand for broadband Internet services in Austria focusing on residential consumers, was conducted by Cardona, Schwarz, Yurtoglu, and Zulehner (2009). Using 3000 households and analyzing four types of Internet access: narrowband, cable, DSL and mobile, they found that demand for DSL is elastic and cable networks are likely to be in the same market as DSL connections.
This study must be contextualized; narrowband was the first to arrive and it is not an option anymore. The three remaining services will strongly depend on the development of the infrastructures. Since this study, Internet services have evolved. In A1, an Internet provider in Austria fixed the typical prices of Internet at speeds of 50MBPS and 100MBPS to 29,90€
and 44,90€ respectively.
Choice of alternative types of Internet access will depend on price, availability, but also the utility that consumers give to this service. Some are willing to pay more for the same service. Using a large sample of individuals, Rosston, Savage, and Waldman (2010) study this phenomenon, comparing experience users to inexperience users. In their sample, 5799 were experience users and 479 inexperience users. The willingness to pay is estimated which is represented by the marginal utility of changing from one service (Internet speed) to other service but with higher speeds. In this context an experienced user is a user that had used Internet more than twelve months. Several measures are included in their analysis; cost,
25 connection speed, reliability, use Internet away from home, watch high definition content, interaction with health specialists and being able to perform free videophone calls over the Internet. An experienced household is willing to pay 59$ for a basic service17, 85$ for a premium service18 and 98$ for a premium plus service19, while an inexperienced user is only willing to pay 31$, 59$ and 71$ respectively for an improvement on these services. These results show that being able to work with Internet will affect its utility and that the willingness to pay for additional services depend as well on his utility.
These numbers reported here cannot be extended to countries in Europe; the willingness to pay in Europe would be far less than the reported by this study. Infrastructures in Europe allow smaller prices and higher speeds. Each country has several Internet providers that cover a small geographical area while the USA has the same geographical area as Europe, which can make difficult the development of infrastructures that allow higher
These numbers reported here cannot be extended to countries in Europe; the willingness to pay in Europe would be far less than the reported by this study. Infrastructures in Europe allow smaller prices and higher speeds. Each country has several Internet providers that cover a small geographical area while the USA has the same geographical area as Europe, which can make difficult the development of infrastructures that allow higher